Many persons with chronic mental illness living in the community are at high behavioral risk for HIV infection due to sexual behavior practices and substance use. Vulnerability to HIV risk behaviors in this population is associated with characteristics of chronic mental illness (often including poor social problem-solving, impulsivity, sexual acting out, and co-existing substance use); social interaction patterns (transient social and sexual relationships, relationships with others who share similar social problems, and vulnerability to exploitation and coercion); and the urban inner-city areas where many chronic mentally ill live and where HIV, STD's, and drug use are prevalent. The intent of this research is to adapt social learning and cognitive behavioral skills training principles already successfully used in other "community living" approaches to the problem of HIV risk behavior reduction for the chronic mentally ill. An initial formative research phase will elicit information on risk attitudes, knowledge, and risk-producing situations encountered by the chronic mentally ill. Then, from the population of an urban mental health clinic, 400 persons (200 men and 200 women) who meet criteria for HIV high-risk behavior will be recruited. Pilot data already collected indicate that risk behavior and STD's are very prevalent in this population. Subjects will be assessed using a battery of measures of risk knowledge, sexual and substance use behavior, cognitive and behavioral skill pertinent to risk reduction implementation, behavioral intentions, and STD incidence. 200 subjects will be randomly assigned to participate in a multiple-session HIV risk reduction group intervention program with followup boosters which includes risk education, behavior change preparedness, intensive behavioral skills training, training in problem-solving skills, and reinforcement of change efforts. The other 200 subjects will participate in a time-matched health intervention program on topics unrelated to AIDS. All participants will be reassessed at post-intervention and through 18-month followup to determine intervention effectiveness. Multivariate analyses will be used to examine short- and long-term change associated with intervention participation. It is hypothesized that intervention participants relative to those in the comparison condition will exhibit increased risk knowledge, improved behavioral and cognitive skill related to change implementation, reduced high-risk behavior, and lower incidence of STD's. The purpose of this work is to develop, implement, and evaluate an AIDS prevention model which can be of practical benefit to health and mental health programs which serve chronic mentally ill persons who engage in behaviors which confer risk for HIV infection and other STD's.